Learning and remembering of 'new' information almost invariably occurs in the presence of a great deal of previously acquired information. A simple logical consequence of this fundamental fact is that adult 'new' learning is frequently a matter of 'relearning' or 'updating' of existing information. Such updating or transfer may or may not follow the same principles as does truly original learning. Relatively little research has been done in cognitive neuroscience of memory on the possible implications of this fact. We propose to do so. Our approach to 'localization of memory processes' differs from conventional ones. While earlier inquiries asked which critical regions are necessary for, or play a role in, initial learning, we believe it is equally important to ask the same question about processes and systems related to the updating existing information and cross-situational transfer of learning. The proposed experiments are designed to delineate the processes and brain structures that are involved in updating existing information. The processes of primary interest are two kinds: (a) those that allow new learning in the form of 'updating' (or temporal sequencing) of information, and (b) those that allow a 'gluing' (or conglomeration) of parts into more complex functional neurocognitive structures. The experiments rely on (a) variations of well-established techniques, such as paired-associate list learning, the testing of serial position effects, A-not-B search tasks, and continuous recognition procedures, and (b) variations of more recently developed paradigms such as errorless learning techniques, fragment recognition, and illusory conjunctions. The stimulus materials consist of words, faces, and difficult-to-name visual patterns. Neuroanatomical bases of the processes of interest will be traced by charting the relative abilities of patients with focal cortical lesions (notably in the dorsolateral prefrontal cortex and the hippocampal regions), severe anterograde amnesics, and control subjects.